xref: /freebsd/sys/netinet/raw_ip.c (revision 5773cccf19ef7b97e56c1101aa481c43149224da)
1 /*
2  * Copyright (c) 1982, 1986, 1988, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. All advertising materials mentioning features or use of this software
14  *    must display the following acknowledgement:
15  *	This product includes software developed by the University of
16  *	California, Berkeley and its contributors.
17  * 4. Neither the name of the University nor the names of its contributors
18  *    may be used to endorse or promote products derived from this software
19  *    without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  *
33  *	@(#)raw_ip.c	8.7 (Berkeley) 5/15/95
34  * $FreeBSD$
35  */
36 
37 #include "opt_inet6.h"
38 #include "opt_ipsec.h"
39 #include "opt_mac.h"
40 #include "opt_random_ip_id.h"
41 
42 #include <sys/param.h>
43 #include <sys/kernel.h>
44 #include <sys/lock.h>
45 #include <sys/mac.h>
46 #include <sys/malloc.h>
47 #include <sys/mbuf.h>
48 #include <sys/proc.h>
49 #include <sys/protosw.h>
50 #include <sys/signalvar.h>
51 #include <sys/socket.h>
52 #include <sys/socketvar.h>
53 #include <sys/sx.h>
54 #include <sys/sysctl.h>
55 #include <sys/systm.h>
56 
57 #include <vm/uma.h>
58 
59 #include <net/if.h>
60 #include <net/route.h>
61 
62 #include <netinet/in.h>
63 #include <netinet/in_systm.h>
64 #include <netinet/in_pcb.h>
65 #include <netinet/in_var.h>
66 #include <netinet/ip.h>
67 #include <netinet/ip_var.h>
68 #include <netinet/ip_mroute.h>
69 
70 #include <netinet/ip_fw.h>
71 #include <netinet/ip_dummynet.h>
72 
73 #ifdef FAST_IPSEC
74 #include <netipsec/ipsec.h>
75 #endif /*FAST_IPSEC*/
76 
77 #ifdef IPSEC
78 #include <netinet6/ipsec.h>
79 #endif /*IPSEC*/
80 
81 struct	inpcbhead ripcb;
82 struct	inpcbinfo ripcbinfo;
83 
84 /* control hooks for ipfw and dummynet */
85 ip_fw_ctl_t *ip_fw_ctl_ptr;
86 ip_dn_ctl_t *ip_dn_ctl_ptr;
87 
88 /*
89  * hooks for multicast routing. They all default to NULL,
90  * so leave them not initialized and rely on BSS being set to 0.
91  */
92 
93 /* The socket used to communicate with the multicast routing daemon.  */
94 struct socket  *ip_mrouter;
95 
96 /* The various mrouter and rsvp functions */
97 int (*ip_mrouter_set)(struct socket *, struct sockopt *);
98 int (*ip_mrouter_get)(struct socket *, struct sockopt *);
99 int (*ip_mrouter_done)(void);
100 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
101                    struct ip_moptions *);
102 int (*mrt_ioctl)(int, caddr_t);
103 int (*legal_vif_num)(int);
104 u_long (*ip_mcast_src)(int);
105 
106 void (*rsvp_input_p)(struct mbuf *m, int off);
107 int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
108 void (*ip_rsvp_force_done)(struct socket *);
109 
110 /*
111  * Nominal space allocated to a raw ip socket.
112  */
113 #define	RIPSNDQ		8192
114 #define	RIPRCVQ		8192
115 
116 /*
117  * Raw interface to IP protocol.
118  */
119 
120 /*
121  * Initialize raw connection block q.
122  */
123 void
124 rip_init()
125 {
126 	INP_INFO_LOCK_INIT(&ripcbinfo, "rip");
127 	LIST_INIT(&ripcb);
128 	ripcbinfo.listhead = &ripcb;
129 	/*
130 	 * XXX We don't use the hash list for raw IP, but it's easier
131 	 * to allocate a one entry hash list than it is to check all
132 	 * over the place for hashbase == NULL.
133 	 */
134 	ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
135 	ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
136 	ripcbinfo.ipi_zone = uma_zcreate("ripcb", sizeof(struct inpcb),
137 	    NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE);
138 	uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets);
139 }
140 
141 static struct	sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
142 /*
143  * Setup generic address and protocol structures
144  * for raw_input routine, then pass them along with
145  * mbuf chain.
146  */
147 void
148 rip_input(m, off)
149 	struct mbuf *m;
150 	int off;
151 {
152 	register struct ip *ip = mtod(m, struct ip *);
153 	register struct inpcb *inp;
154 	struct inpcb *last = 0;
155 	struct mbuf *opts = 0;
156 	int proto = ip->ip_p;
157 
158 	ripsrc.sin_addr = ip->ip_src;
159 	LIST_FOREACH(inp, &ripcb, inp_list) {
160 #ifdef INET6
161 		if ((inp->inp_vflag & INP_IPV4) == 0)
162 			continue;
163 #endif
164 		if (inp->inp_ip_p && inp->inp_ip_p != proto)
165 			continue;
166 		if (inp->inp_laddr.s_addr &&
167                   inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
168 			continue;
169 		if (inp->inp_faddr.s_addr &&
170                   inp->inp_faddr.s_addr != ip->ip_src.s_addr)
171 			continue;
172 		if (last) {
173 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
174 			int policyfail = 0;
175 
176 			if (n != NULL) {
177 #ifdef IPSSEC
178 				/* check AH/ESP integrity. */
179 				if (ipsec4_in_reject_so(n, last->inp_socket)) {
180 					policyfail = 1;
181 					ipsecstat.in_polvio++;
182 					/* do not inject data to pcb */
183 				}
184 #endif /*IPSEC*/
185 #ifdef FAST_IPSEC
186 				/* check AH/ESP integrity. */
187 				if (ipsec4_in_reject(n, last)) {
188 					policyfail = 1;
189 					/* do not inject data to pcb */
190 				}
191 #endif /*FAST_IPSEC*/
192 #ifdef MAC
193 				if (policyfail == 0 &&
194 				    mac_check_socket_deliver(last->inp_socket,
195 				    n) != 0)
196 					policyfail = 1;
197 #endif
198 			}
199 			if (policyfail)
200 				m_freem(n);
201 			else if (n) {
202 				if (last->inp_flags & INP_CONTROLOPTS ||
203 				    last->inp_socket->so_options & SO_TIMESTAMP)
204 				    ip_savecontrol(last, &opts, ip, n);
205 				if (sbappendaddr(&last->inp_socket->so_rcv,
206 				    (struct sockaddr *)&ripsrc, n,
207 				    opts) == 0) {
208 					/* should notify about lost packet */
209 					m_freem(n);
210 					if (opts)
211 					    m_freem(opts);
212 				} else
213 					sorwakeup(last->inp_socket);
214 				opts = 0;
215 			}
216 		}
217 		last = inp;
218 	}
219 	if (last) {
220 #ifdef IPSEC
221 		/* check AH/ESP integrity. */
222 		if (ipsec4_in_reject_so(m, last->inp_socket)) {
223 			m_freem(m);
224 			ipsecstat.in_polvio++;
225 			ipstat.ips_delivered--;
226 			/* do not inject data to pcb */
227 			return;
228 		}
229 #endif /*IPSEC*/
230 #ifdef FAST_IPSEC
231 		/* check AH/ESP integrity. */
232 		if (ipsec4_in_reject(m, last)) {
233 			m_freem(m);
234 			ipstat.ips_delivered--;
235 			/* do not inject data to pcb */
236 			return;
237 		}
238 #endif /*FAST_IPSEC*/
239 #ifdef MAC
240 		if (mac_check_socket_deliver(last->inp_socket, m) != 0) {
241 			m_freem(m);
242 			ipstat.ips_delivered--;
243 			return;
244 		}
245 #endif
246 		if (last->inp_flags & INP_CONTROLOPTS ||
247 		    last->inp_socket->so_options & SO_TIMESTAMP)
248 			ip_savecontrol(last, &opts, ip, m);
249 		if (sbappendaddr(&last->inp_socket->so_rcv,
250 		    (struct sockaddr *)&ripsrc, m, opts) == 0) {
251 			m_freem(m);
252 			if (opts)
253 			    m_freem(opts);
254 		} else
255 			sorwakeup(last->inp_socket);
256 	} else {
257 		m_freem(m);
258 		ipstat.ips_noproto++;
259 		ipstat.ips_delivered--;
260 	}
261 }
262 
263 /*
264  * Generate IP header and pass packet to ip_output.
265  * Tack on options user may have setup with control call.
266  */
267 int
268 rip_output(m, so, dst)
269 	struct mbuf *m;
270 	struct socket *so;
271 	u_long dst;
272 {
273 	register struct ip *ip;
274 	register struct inpcb *inp = sotoinpcb(so);
275 	int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
276 
277 #ifdef MAC
278 	mac_create_mbuf_from_socket(so, m);
279 #endif
280 
281 	/*
282 	 * If the user handed us a complete IP packet, use it.
283 	 * Otherwise, allocate an mbuf for a header and fill it in.
284 	 */
285 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
286 		if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
287 			m_freem(m);
288 			return(EMSGSIZE);
289 		}
290 		M_PREPEND(m, sizeof(struct ip), M_TRYWAIT);
291 		ip = mtod(m, struct ip *);
292 		ip->ip_tos = inp->inp_ip_tos;
293 		ip->ip_off = 0;
294 		ip->ip_p = inp->inp_ip_p;
295 		ip->ip_len = m->m_pkthdr.len;
296 		ip->ip_src = inp->inp_laddr;
297 		ip->ip_dst.s_addr = dst;
298 		ip->ip_ttl = inp->inp_ip_ttl;
299 	} else {
300 		if (m->m_pkthdr.len > IP_MAXPACKET) {
301 			m_freem(m);
302 			return(EMSGSIZE);
303 		}
304 		ip = mtod(m, struct ip *);
305 		/* don't allow both user specified and setsockopt options,
306 		   and don't allow packet length sizes that will crash */
307 		if (((ip->ip_hl != (sizeof (*ip) >> 2))
308 		     && inp->inp_options)
309 		    || (ip->ip_len > m->m_pkthdr.len)
310 		    || (ip->ip_len < (ip->ip_hl << 2))) {
311 			m_freem(m);
312 			return EINVAL;
313 		}
314 		if (ip->ip_id == 0)
315 #ifdef RANDOM_IP_ID
316 			ip->ip_id = ip_randomid();
317 #else
318 			ip->ip_id = htons(ip_id++);
319 #endif
320 		/* XXX prevent ip_output from overwriting header fields */
321 		flags |= IP_RAWOUTPUT;
322 		ipstat.ips_rawout++;
323 	}
324 
325 	return (ip_output(m, inp->inp_options, &inp->inp_route, flags,
326 			  inp->inp_moptions, inp));
327 }
328 
329 /*
330  * Raw IP socket option processing.
331  */
332 int
333 rip_ctloutput(so, sopt)
334 	struct socket *so;
335 	struct sockopt *sopt;
336 {
337 	struct	inpcb *inp = sotoinpcb(so);
338 	int	error, optval;
339 
340 	if (sopt->sopt_level != IPPROTO_IP)
341 		return (EINVAL);
342 
343 	error = 0;
344 
345 	switch (sopt->sopt_dir) {
346 	case SOPT_GET:
347 		switch (sopt->sopt_name) {
348 		case IP_HDRINCL:
349 			optval = inp->inp_flags & INP_HDRINCL;
350 			error = sooptcopyout(sopt, &optval, sizeof optval);
351 			break;
352 
353 		case IP_FW_ADD:	/* ADD actually returns the body... */
354 		case IP_FW_GET:
355 			if (IPFW_LOADED)
356 				error = ip_fw_ctl_ptr(sopt);
357 			else
358 				error = ENOPROTOOPT;
359 			break;
360 
361 		case IP_DUMMYNET_GET:
362 			if (DUMMYNET_LOADED)
363 				error = ip_dn_ctl_ptr(sopt);
364 			else
365 				error = ENOPROTOOPT;
366 			break ;
367 
368 		case MRT_INIT:
369 		case MRT_DONE:
370 		case MRT_ADD_VIF:
371 		case MRT_DEL_VIF:
372 		case MRT_ADD_MFC:
373 		case MRT_DEL_MFC:
374 		case MRT_VERSION:
375 		case MRT_ASSERT:
376 			error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
377 				EOPNOTSUPP;
378 			break;
379 
380 		default:
381 			error = ip_ctloutput(so, sopt);
382 			break;
383 		}
384 		break;
385 
386 	case SOPT_SET:
387 		switch (sopt->sopt_name) {
388 		case IP_HDRINCL:
389 			error = sooptcopyin(sopt, &optval, sizeof optval,
390 					    sizeof optval);
391 			if (error)
392 				break;
393 			if (optval)
394 				inp->inp_flags |= INP_HDRINCL;
395 			else
396 				inp->inp_flags &= ~INP_HDRINCL;
397 			break;
398 
399 		case IP_FW_ADD:
400 		case IP_FW_DEL:
401 		case IP_FW_FLUSH:
402 		case IP_FW_ZERO:
403 		case IP_FW_RESETLOG:
404 			if (IPFW_LOADED)
405 				error = ip_fw_ctl_ptr(sopt);
406 			else
407 				error = ENOPROTOOPT;
408 			break;
409 
410 		case IP_DUMMYNET_CONFIGURE:
411 		case IP_DUMMYNET_DEL:
412 		case IP_DUMMYNET_FLUSH:
413 			if (DUMMYNET_LOADED)
414 				error = ip_dn_ctl_ptr(sopt);
415 			else
416 				error = ENOPROTOOPT ;
417 			break ;
418 
419 		case IP_RSVP_ON:
420 			error = ip_rsvp_init(so);
421 			break;
422 
423 		case IP_RSVP_OFF:
424 			error = ip_rsvp_done();
425 			break;
426 
427 		case IP_RSVP_VIF_ON:
428 		case IP_RSVP_VIF_OFF:
429 			error = ip_rsvp_vif ?
430 				ip_rsvp_vif(so, sopt) : EINVAL;
431 			break;
432 
433 		case MRT_INIT:
434 		case MRT_DONE:
435 		case MRT_ADD_VIF:
436 		case MRT_DEL_VIF:
437 		case MRT_ADD_MFC:
438 		case MRT_DEL_MFC:
439 		case MRT_VERSION:
440 		case MRT_ASSERT:
441 			error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
442 					EOPNOTSUPP;
443 			break;
444 
445 		default:
446 			error = ip_ctloutput(so, sopt);
447 			break;
448 		}
449 		break;
450 	}
451 
452 	return (error);
453 }
454 
455 /*
456  * This function exists solely to receive the PRC_IFDOWN messages which
457  * are sent by if_down().  It looks for an ifaddr whose ifa_addr is sa,
458  * and calls in_ifadown() to remove all routes corresponding to that address.
459  * It also receives the PRC_IFUP messages from if_up() and reinstalls the
460  * interface routes.
461  */
462 void
463 rip_ctlinput(cmd, sa, vip)
464 	int cmd;
465 	struct sockaddr *sa;
466 	void *vip;
467 {
468 	struct in_ifaddr *ia;
469 	struct ifnet *ifp;
470 	int err;
471 	int flags;
472 
473 	switch (cmd) {
474 	case PRC_IFDOWN:
475 		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
476 			if (ia->ia_ifa.ifa_addr == sa
477 			    && (ia->ia_flags & IFA_ROUTE)) {
478 				/*
479 				 * in_ifscrub kills the interface route.
480 				 */
481 				in_ifscrub(ia->ia_ifp, ia);
482 				/*
483 				 * in_ifadown gets rid of all the rest of
484 				 * the routes.  This is not quite the right
485 				 * thing to do, but at least if we are running
486 				 * a routing process they will come back.
487 				 */
488 				in_ifadown(&ia->ia_ifa, 0);
489 				break;
490 			}
491 		}
492 		break;
493 
494 	case PRC_IFUP:
495 		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
496 			if (ia->ia_ifa.ifa_addr == sa)
497 				break;
498 		}
499 		if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
500 			return;
501 		flags = RTF_UP;
502 		ifp = ia->ia_ifa.ifa_ifp;
503 
504 		if ((ifp->if_flags & IFF_LOOPBACK)
505 		    || (ifp->if_flags & IFF_POINTOPOINT))
506 			flags |= RTF_HOST;
507 
508 		err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
509 		if (err == 0)
510 			ia->ia_flags |= IFA_ROUTE;
511 		break;
512 	}
513 }
514 
515 u_long	rip_sendspace = RIPSNDQ;
516 u_long	rip_recvspace = RIPRCVQ;
517 int	rip_olddiverterror = 1;
518 
519 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
520     &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
521 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
522     &rip_recvspace, 0, "Maximum incoming raw IP datagram size");
523 SYSCTL_INT(_net_inet_raw, OID_AUTO, olddiverterror, CTLFLAG_RW,
524     &rip_olddiverterror, 0, "Return an error when creating an 'old' DIVERT socket");
525 
526 static int
527 rip_attach(struct socket *so, int proto, struct thread *td)
528 {
529 	struct inpcb *inp;
530 	int error, s;
531 
532 	inp = sotoinpcb(so);
533 	if (inp)
534 		panic("rip_attach");
535 	if (td && (error = suser(td)) != 0)
536 		return error;
537 
538 	if (proto >= IPPROTO_MAX || proto < 0)
539 		return EPROTONOSUPPORT;
540 
541 	/* To be removed before 5.2 */
542 	if (rip_olddiverterror && proto == IPPROTO_OLD_DIVERT) {
543 		printf("Old IPDIVERT program needs to be recompiled, or new IP proto 254 user needs sysctl net.inet.raw.olddiverterror=0\n");
544 		return EPROTONOSUPPORT;
545 	}
546 
547 	error = soreserve(so, rip_sendspace, rip_recvspace);
548 	if (error)
549 		return error;
550 	s = splnet();
551 	error = in_pcballoc(so, &ripcbinfo, td);
552 	splx(s);
553 	if (error)
554 		return error;
555 	inp = (struct inpcb *)so->so_pcb;
556 	inp->inp_vflag |= INP_IPV4;
557 	inp->inp_ip_p = proto;
558 	inp->inp_ip_ttl = ip_defttl;
559 	return 0;
560 }
561 
562 static int
563 rip_detach(struct socket *so)
564 {
565 	struct inpcb *inp;
566 
567 	inp = sotoinpcb(so);
568 	if (inp == 0)
569 		panic("rip_detach");
570 	if (so == ip_mrouter && ip_mrouter_done)
571 		ip_mrouter_done();
572 	if (ip_rsvp_force_done)
573 		ip_rsvp_force_done(so);
574 	if (so == ip_rsvpd)
575 		ip_rsvp_done();
576 	in_pcbdetach(inp);
577 	return 0;
578 }
579 
580 static int
581 rip_abort(struct socket *so)
582 {
583 	soisdisconnected(so);
584 	return rip_detach(so);
585 }
586 
587 static int
588 rip_disconnect(struct socket *so)
589 {
590 	if ((so->so_state & SS_ISCONNECTED) == 0)
591 		return ENOTCONN;
592 	return rip_abort(so);
593 }
594 
595 static int
596 rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
597 {
598 	struct inpcb *inp = sotoinpcb(so);
599 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
600 
601 	if (nam->sa_len != sizeof(*addr))
602 		return EINVAL;
603 
604 	if (TAILQ_EMPTY(&ifnet) || ((addr->sin_family != AF_INET) &&
605 				    (addr->sin_family != AF_IMPLINK)) ||
606 	    (addr->sin_addr.s_addr &&
607 	     ifa_ifwithaddr((struct sockaddr *)addr) == 0))
608 		return EADDRNOTAVAIL;
609 	inp->inp_laddr = addr->sin_addr;
610 	return 0;
611 }
612 
613 static int
614 rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
615 {
616 	struct inpcb *inp = sotoinpcb(so);
617 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
618 
619 	if (nam->sa_len != sizeof(*addr))
620 		return EINVAL;
621 	if (TAILQ_EMPTY(&ifnet))
622 		return EADDRNOTAVAIL;
623 	if ((addr->sin_family != AF_INET) &&
624 	    (addr->sin_family != AF_IMPLINK))
625 		return EAFNOSUPPORT;
626 	inp->inp_faddr = addr->sin_addr;
627 	soisconnected(so);
628 	return 0;
629 }
630 
631 static int
632 rip_shutdown(struct socket *so)
633 {
634 	socantsendmore(so);
635 	return 0;
636 }
637 
638 static int
639 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
640 	 struct mbuf *control, struct thread *td)
641 {
642 	struct inpcb *inp = sotoinpcb(so);
643 	register u_long dst;
644 
645 	if (so->so_state & SS_ISCONNECTED) {
646 		if (nam) {
647 			m_freem(m);
648 			return EISCONN;
649 		}
650 		dst = inp->inp_faddr.s_addr;
651 	} else {
652 		if (nam == NULL) {
653 			m_freem(m);
654 			return ENOTCONN;
655 		}
656 		dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
657 	}
658 	return rip_output(m, so, dst);
659 }
660 
661 static int
662 rip_pcblist(SYSCTL_HANDLER_ARGS)
663 {
664 	int error, i, n, s;
665 	struct inpcb *inp, **inp_list;
666 	inp_gen_t gencnt;
667 	struct xinpgen xig;
668 
669 	/*
670 	 * The process of preparing the TCB list is too time-consuming and
671 	 * resource-intensive to repeat twice on every request.
672 	 */
673 	if (req->oldptr == 0) {
674 		n = ripcbinfo.ipi_count;
675 		req->oldidx = 2 * (sizeof xig)
676 			+ (n + n/8) * sizeof(struct xinpcb);
677 		return 0;
678 	}
679 
680 	if (req->newptr != 0)
681 		return EPERM;
682 
683 	/*
684 	 * OK, now we're committed to doing something.
685 	 */
686 	s = splnet();
687 	gencnt = ripcbinfo.ipi_gencnt;
688 	n = ripcbinfo.ipi_count;
689 	splx(s);
690 
691 	xig.xig_len = sizeof xig;
692 	xig.xig_count = n;
693 	xig.xig_gen = gencnt;
694 	xig.xig_sogen = so_gencnt;
695 	error = SYSCTL_OUT(req, &xig, sizeof xig);
696 	if (error)
697 		return error;
698 
699 	inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
700 	if (inp_list == 0)
701 		return ENOMEM;
702 
703 	s = splnet();
704 	for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n;
705 	     inp = LIST_NEXT(inp, inp_list)) {
706 		if (inp->inp_gencnt <= gencnt) {
707 			if (cr_canseesocket(req->td->td_ucred,
708 			    inp->inp_socket))
709 				continue;
710 			inp_list[i++] = inp;
711 		}
712 	}
713 	splx(s);
714 	n = i;
715 
716 	error = 0;
717 	for (i = 0; i < n; i++) {
718 		inp = inp_list[i];
719 		if (inp->inp_gencnt <= gencnt) {
720 			struct xinpcb xi;
721 			xi.xi_len = sizeof xi;
722 			/* XXX should avoid extra copy */
723 			bcopy(inp, &xi.xi_inp, sizeof *inp);
724 			if (inp->inp_socket)
725 				sotoxsocket(inp->inp_socket, &xi.xi_socket);
726 			error = SYSCTL_OUT(req, &xi, sizeof xi);
727 		}
728 	}
729 	if (!error) {
730 		/*
731 		 * Give the user an updated idea of our state.
732 		 * If the generation differs from what we told
733 		 * her before, she knows that something happened
734 		 * while we were processing this request, and it
735 		 * might be necessary to retry.
736 		 */
737 		s = splnet();
738 		xig.xig_gen = ripcbinfo.ipi_gencnt;
739 		xig.xig_sogen = so_gencnt;
740 		xig.xig_count = ripcbinfo.ipi_count;
741 		splx(s);
742 		error = SYSCTL_OUT(req, &xig, sizeof xig);
743 	}
744 	free(inp_list, M_TEMP);
745 	return error;
746 }
747 
748 /*
749  * This is the wrapper function for in_setsockaddr.  We just pass down
750  * the pcbinfo for in_setpeeraddr to lock.
751  */
752 static int
753 rip_sockaddr(struct socket *so, struct sockaddr **nam)
754 {
755 	return (in_setsockaddr(so, nam, &ripcbinfo));
756 }
757 
758 /*
759  * This is the wrapper function for in_setpeeraddr.  We just pass down
760  * the pcbinfo for in_setpeeraddr to lock.
761  */
762 static int
763 rip_peeraddr(struct socket *so, struct sockaddr **nam)
764 {
765 	return (in_setpeeraddr(so, nam, &ripcbinfo));
766 }
767 
768 
769 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
770 	    rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
771 
772 struct pr_usrreqs rip_usrreqs = {
773 	rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect,
774 	pru_connect2_notsupp, in_control, rip_detach, rip_disconnect,
775 	pru_listen_notsupp, rip_peeraddr, pru_rcvd_notsupp,
776 	pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown,
777 	rip_sockaddr, sosend, soreceive, sopoll
778 };
779